There is considerable interest in the potential effect on benign disease of a diet free of methylxanthines (caffeine, theophylline, and theobromine) found in coffee, tea, colas, and chocolate. We randomly assigned 158 women who presented with a breast concern either to a group encouraged to abstain from methylxanthine-containing foods and beverages or to a group who received no dietary recommendations (controls). At the initial visit each patient's sociomedical history and data on methylxanthine consumption were obtained by interview, and clinically palpable breast findings were graded on a scale of 0 to 4 (no nodularity to confluent hard "dysplasia") for each quadrant of both breasts. On the follow-up visit approximately 4 months later similar information was obtained. Mammograms were taken at both visits for a subset of women in each group. We found a statistically significant reduction in clinically palpable breast findings in the abstaining group as compared with controls, but the absolute change was minor and may be of little clinical significance. Comparison of before-after mammograms offered little support for the methylxanthine hypothesis. There was no relation between clinically palpable breast finding scores at initial examination and caffeine consumption levels reported at that time.